Probability of extreme hydrometeorological events - a different approach

A criticism of the standard approach to frequency analysis is presented and a different rationale is proposed based on treating the variable of interest as a compound event and synthesizing its distribution function by enumerating all physically plausible combinations of its major components for which data are available. Since only the order of magnitude of the probabilities of the extreme events is aimed for, the fine points of small sample theory, in which the standard approach is entangled, become spurious while more room is made available for explicit consideration of the physical processes involved. The approach is illustrated by an example using data on annual maxima of daily precipitation from the coastal region of British Columbia.